Fuzzy Real-time Advanced Shell for Intelligent Control with Fuzzy Algorithm Compiler

Abstract

In general, an expert system which supports the conventional fuzzy reasoning has relatively high user interface functions like the rule editor by multi-windows and others. However, it has suffered from some problems, because it has a hard time linking with a user program as it is rule-driven on a shell or because its reasoning speed is slow, as the shell system tends to be huge, and it is, therefore, not suited to real-time control. Under these circumstances, the present author and others have developed a user-program-initiated fuzzy shell FRASH (Fuzzy Real-time Advanced SHell) which is used for intelligent control in real time. This paper is intended mainly to describe the basic configuration of FRASH and its expression functions. This shell has a number of features, such as a high-speed reasoning engine in the form of a library, a function for online-tuning of all the data in the reasoning, an off-line editor function based on multi-windows, a function for displaying reasoning states, a capability to express fuzzy frames which can define fuzzy numbers as slot values, a fuzzy algorithm compiler based on macro-descriptions, and so on. In addition, with a view to confirming the basic functions of the shell, a simple simulation was carried out by taking up vehicle tracking and overtaking control as an example. By this simulation, it was demonstrated that the fuzzy reasoning function of FRASH is sufficiently fast for mechatronic control in real time and that the fuzzy frame and fuzzy algorithm functions are effective in describing macrolevel decision-making problems.